This is the Official implementation for "PEAC: Unsupervised Pre-training for Cross-Embodiment Reinforcement Learning" (NeurIPS 2024)
The code is based on URLB
You can create an anaconda environment and install all required dependencies by running
conda create -n ceurl python=3.8 conda activate ceurl pip install -r requirements.txt pip install torch==1.8.0+cu111 torchvision==0.9.0+cu111 -f https://download.pytorch.org/whl/torch_stable.htmlThe simplest way to try PEAC in three embodiment distributions of state-based DMC by running
cd DMC_state chmod +x train_finetune.sh ./train_finetune.sh peac walker_mass 0 ./train_finetune.sh peac quadruped_mass 0 ./train_finetune.sh peac quadruped_damping 0The simplest way to try PEAC in three embodiment distributions of image-based DMC by running
cd DMC_image chmod +x train_finetune.sh ./train_finetune.sh peac_lbs walker_mass 0 ./train_finetune.sh peac_lbs quadruped_mass 0 ./train_finetune.sh peac_lbs quadruped_damping 0 ./train_finetune.sh peac_diayn walker_mass 0 ./train_finetune.sh peac_diayn quadruped_mass 0 ./train_finetune.sh peac_diayn quadruped_damping 0If you find this work helpful, please cite our paper.
@article{ying2024peac, title={PEAC: Unsupervised Pre-training for Cross-Embodiment Reinforcement Learning}, author={Ying, Chengyang and Hao, Zhongkai and Zhou, Xinning and Xu, Xuezhou and Su, Hang and Zhang, Xingxing and Zhu, Jun}, journal={arXiv preprint arXiv:2405.14073}, year={2024} }